David Brown | fecda2d | 2017-09-07 10:20:34 -0600 | [diff] [blame] | 1 | /* ecc.c - TinyCrypt implementation of ECC auxiliary functions */ |
| 2 | |
| 3 | /* |
| 4 | * |
| 5 | * Copyright (c) 2013, Kenneth MacKay |
| 6 | * All rights reserved. |
| 7 | * https://github.com/kmackay/micro-ecc |
| 8 | * |
| 9 | * Redistribution and use in source and binary forms, with or without modification, |
| 10 | * are permitted provided that the following conditions are met: |
| 11 | * * Redistributions of source code must retain the above copyright notice, this |
| 12 | * list of conditions and the following disclaimer. |
| 13 | * * Redistributions in binary form must reproduce the above copyright notice, |
| 14 | * this list of conditions and the following disclaimer in the documentation |
| 15 | * and/or other materials provided with the distribution. |
| 16 | * |
| 17 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND |
| 18 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED |
| 19 | * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE |
| 20 | * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR |
| 21 | * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
| 22 | * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
| 23 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON |
| 24 | * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 25 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS |
| 26 | * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 27 | * |
| 28 | * Copyright (C) 2015 by Intel Corporation, All Rights Reserved. |
| 29 | * |
| 30 | * Redistribution and use in source and binary forms, with or without |
| 31 | * modification, are permitted provided that the following conditions are met: |
| 32 | * |
| 33 | * - Redistributions of source code must retain the above copyright notice, |
| 34 | * this list of conditions and the following disclaimer. |
| 35 | * |
| 36 | * - Redistributions in binary form must reproduce the above copyright |
| 37 | * notice, this list of conditions and the following disclaimer in the |
| 38 | * documentation and/or other materials provided with the distribution. |
| 39 | * |
| 40 | * - Neither the name of Intel Corporation nor the names of its contributors |
| 41 | * may be used to endorse or promote products derived from this software |
| 42 | * without specific prior written permission. |
| 43 | * |
| 44 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
| 45 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 46 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 47 | * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE |
| 48 | * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 49 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 50 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 51 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 52 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 53 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| 54 | * POSSIBILITY OF SUCH DAMAGE. |
| 55 | */ |
| 56 | |
| 57 | #include <tinycrypt/ecc.h> |
| 58 | |
| 59 | /* ------ Curve NIST P-256 constants: ------ */ |
| 60 | |
| 61 | #define Curve_P {0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, \ |
| 62 | 0x00000000, 0x00000000, 0x00000001, 0xFFFFFFFF} |
| 63 | |
| 64 | #define Curve_B {0x27D2604B, 0x3BCE3C3E, 0xCC53B0F6, 0x651D06B0, \ |
| 65 | 0x769886BC, 0xB3EBBD55, 0xAA3A93E7, 0x5AC635D8} |
| 66 | |
| 67 | #define Curve_N {0xFC632551, 0xF3B9CAC2, 0xA7179E84, 0xBCE6FAAD, \ |
| 68 | 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, 0xFFFFFFFF} |
| 69 | |
| 70 | #define Curve_G {{0xD898C296, 0xF4A13945, 0x2DEB33A0, 0x77037D81, \ |
| 71 | 0x63A440F2, 0xF8BCE6E5, 0xE12C4247, 0x6B17D1F2}, \ |
| 72 | {0x37BF51F5, 0xCBB64068, 0x6B315ECE, 0x2BCE3357, \ |
| 73 | 0x7C0F9E16, 0x8EE7EB4A, 0xFE1A7F9B, 0x4FE342E2} } |
| 74 | |
| 75 | #define Curve_P_Barrett {0x00000003, 0x00000000, 0xFFFFFFFF, 0xFFFFFFFE, \ |
| 76 | 0xFFFFFFFE, 0xFFFFFFFE, 0xFFFFFFFF, 0x00000000, 0x00000001} |
| 77 | |
| 78 | #define Curve_N_Barrett {0xEEDF9BFE, 0x012FFD85, 0xDF1A6C21, 0x43190552, \ |
| 79 | 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF, 0x00000000, 0x00000001} |
| 80 | |
| 81 | uint32_t curve_p[NUM_ECC_DIGITS] = Curve_P; |
| 82 | uint32_t curve_b[NUM_ECC_DIGITS] = Curve_B; |
| 83 | EccPoint curve_G = Curve_G; |
| 84 | uint32_t curve_n[NUM_ECC_DIGITS] = Curve_N; |
| 85 | uint32_t curve_pb[NUM_ECC_DIGITS + 1] = Curve_P_Barrett; |
| 86 | uint32_t curve_nb[NUM_ECC_DIGITS + 1] = Curve_N_Barrett; |
| 87 | |
| 88 | /* ------ Static functions: ------ */ |
| 89 | |
| 90 | /* Zeroing out p_vli. */ |
| 91 | static void vli_clear(uint32_t *p_vli) |
| 92 | { |
| 93 | uint32_t i; |
| 94 | |
| 95 | for (i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 96 | p_vli[i] = 0; |
| 97 | } |
| 98 | } |
| 99 | |
| 100 | /* Returns nonzero if bit p_bit of p_vli is set. |
| 101 | * It is assumed that the value provided in 'bit' is within |
| 102 | * the boundaries of the word-array 'p_vli'.*/ |
| 103 | static uint32_t vli_testBit(uint32_t *p_vli, uint32_t p_bit) |
| 104 | { |
| 105 | return (p_vli[p_bit / 32] & (1 << (p_bit % 32))); |
| 106 | } |
| 107 | |
| 108 | uint32_t vli_isZero(uint32_t *p_vli) |
| 109 | { |
| 110 | uint32_t acc = 0; |
| 111 | |
| 112 | for (uint32_t i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 113 | acc |= p_vli[i]; |
| 114 | } |
| 115 | |
| 116 | return (!acc); |
| 117 | } |
| 118 | |
| 119 | /* |
| 120 | * Find the right-most nonzero 32-bit "digits" in p_vli. |
| 121 | * |
| 122 | * Side-channel countermeasure: algorithm strengthened against timing attack. |
| 123 | */ |
| 124 | static uint32_t vli_numDigits(uint32_t *p_vli) |
| 125 | { |
| 126 | int32_t i; |
| 127 | uint32_t digits = 0; |
| 128 | |
| 129 | for (i = NUM_ECC_DIGITS - 1; i >= 0 ; --i) { |
| 130 | digits += p_vli[i] || digits; |
| 131 | } |
| 132 | |
| 133 | return digits; |
| 134 | } |
| 135 | |
| 136 | /* |
| 137 | * Find the left-most non-zero bit in p_vli. |
| 138 | * |
| 139 | * Side-channel countermeasure: algorithm strengthened against timing attack. |
| 140 | */ |
| 141 | static uint32_t vli_numBits(uint32_t *p_vli) |
| 142 | { |
| 143 | uint32_t l_digit; |
| 144 | uint32_t i, acc = 32; |
| 145 | uint32_t l_numDigits = vli_numDigits(p_vli); |
| 146 | |
| 147 | l_digit = p_vli[l_numDigits - 1]; |
| 148 | |
| 149 | for (i = 0; i < 32; ++i) { |
| 150 | acc -= !l_digit; |
| 151 | l_digit >>= 1; |
| 152 | } |
| 153 | |
| 154 | return ((l_numDigits - 1) * 32 + acc); |
| 155 | } |
| 156 | |
| 157 | /* |
| 158 | * Computes p_result = p_left + p_right, returns carry. |
| 159 | * |
| 160 | * Side-channel countermeasure: algorithm strengthened against timing attack. |
| 161 | */ |
| 162 | static uint32_t vli_add(uint32_t *p_result, uint32_t *p_left, |
| 163 | uint32_t *p_right) |
| 164 | { |
| 165 | |
| 166 | uint32_t l_carry = 0; |
| 167 | |
| 168 | for (uint32_t i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 169 | uint32_t l_sum = p_left[i] + p_right[i] + l_carry; |
| 170 | |
| 171 | l_carry = (l_sum < p_left[i]) | ((l_sum == p_left[i]) && l_carry); |
| 172 | p_result[i] = l_sum; |
| 173 | } |
| 174 | |
| 175 | return l_carry; |
| 176 | } |
| 177 | |
| 178 | |
| 179 | /* Computes p_result = p_left * p_right. */ |
| 180 | static void vli_mult(uint32_t *p_result, uint32_t *p_left, |
| 181 | uint32_t *p_right, uint32_t word_size) |
| 182 | { |
| 183 | |
| 184 | uint64_t r01 = 0; |
| 185 | uint32_t r2 = 0; |
| 186 | |
| 187 | /* Compute each digit of p_result in sequence, maintaining the carries. */ |
| 188 | for (uint32_t k = 0; k < word_size*2 - 1; ++k) { |
| 189 | |
| 190 | uint32_t l_min = (k < word_size ? 0 : (k + 1) - word_size); |
| 191 | |
| 192 | for (uint32_t i = l_min; i <= k && i < word_size; ++i) { |
| 193 | |
| 194 | uint64_t l_product = (uint64_t)p_left[i] * p_right[k - i]; |
| 195 | |
| 196 | r01 += l_product; |
| 197 | r2 += (r01 < l_product); |
| 198 | } |
| 199 | p_result[k] = (uint32_t)r01; |
| 200 | r01 = (r01 >> 32) | (((uint64_t)r2) << 32); |
| 201 | r2 = 0; |
| 202 | } |
| 203 | |
| 204 | p_result[word_size * 2 - 1] = (uint32_t)r01; |
| 205 | } |
| 206 | |
| 207 | /* Computes p_result = p_left^2. */ |
| 208 | static void vli_square(uint32_t *p_result, uint32_t *p_left) |
| 209 | { |
| 210 | |
| 211 | uint64_t r01 = 0; |
| 212 | uint32_t r2 = 0; |
| 213 | uint32_t i, k; |
| 214 | |
| 215 | for (k = 0; k < NUM_ECC_DIGITS * 2 - 1; ++k) { |
| 216 | |
| 217 | uint32_t l_min = (k < NUM_ECC_DIGITS ? 0 : (k + 1) - NUM_ECC_DIGITS); |
| 218 | |
| 219 | for (i = l_min; i <= k && i <= k - i; ++i) { |
| 220 | |
| 221 | uint64_t l_product = (uint64_t)p_left[i] * p_left[k - i]; |
| 222 | |
| 223 | if (i < k - i) { |
| 224 | |
| 225 | r2 += l_product >> 63; |
| 226 | l_product *= 2; |
| 227 | } |
| 228 | r01 += l_product; |
| 229 | r2 += (r01 < l_product); |
| 230 | } |
| 231 | p_result[k] = (uint32_t)r01; |
| 232 | r01 = (r01 >> 32) | (((uint64_t)r2) << 32); |
| 233 | r2 = 0; |
| 234 | } |
| 235 | |
| 236 | p_result[NUM_ECC_DIGITS * 2 - 1] = (uint32_t)r01; |
| 237 | } |
| 238 | |
| 239 | /* Computes p_result = p_product % curve_p using Barrett reduction. */ |
| 240 | void vli_mmod_barrett(uint32_t *p_result, uint32_t *p_product, |
| 241 | uint32_t *p_mod, uint32_t *p_barrett) |
| 242 | { |
| 243 | uint32_t i; |
| 244 | uint32_t q1[NUM_ECC_DIGITS + 1]; |
| 245 | |
| 246 | for (i = NUM_ECC_DIGITS - 1; i < 2 * NUM_ECC_DIGITS; i++) { |
| 247 | q1[i - (NUM_ECC_DIGITS - 1)] = p_product[i]; |
| 248 | } |
| 249 | |
| 250 | uint32_t q2[2*NUM_ECC_DIGITS + 2]; |
| 251 | |
| 252 | vli_mult(q2, q1, p_barrett, NUM_ECC_DIGITS + 1); |
| 253 | for (i = NUM_ECC_DIGITS + 1; i < 2 * NUM_ECC_DIGITS + 2; i++) { |
| 254 | q1[i - (NUM_ECC_DIGITS + 1)] = q2[i]; |
| 255 | } |
| 256 | |
| 257 | uint32_t prime2[2*NUM_ECC_DIGITS]; |
| 258 | |
| 259 | for (i = 0; i < NUM_ECC_DIGITS; i++) { |
| 260 | prime2[i] = p_mod[i]; |
| 261 | prime2[NUM_ECC_DIGITS + i] = 0; |
| 262 | } |
| 263 | |
| 264 | vli_mult(q2, q1, prime2, NUM_ECC_DIGITS + 1); |
| 265 | vli_sub(p_product, p_product, q2, 2 * NUM_ECC_DIGITS); |
| 266 | |
| 267 | uint32_t borrow; |
| 268 | |
| 269 | borrow = vli_sub(q1, p_product, prime2, NUM_ECC_DIGITS + 1); |
| 270 | vli_cond_set(p_product, p_product, q1, borrow); |
| 271 | p_product[NUM_ECC_DIGITS] = q1[NUM_ECC_DIGITS] * (!borrow); |
| 272 | borrow = vli_sub(q1, p_product, prime2, NUM_ECC_DIGITS + 1); |
| 273 | vli_cond_set(p_product, p_product, q1, borrow); |
| 274 | p_product[NUM_ECC_DIGITS] = q1[NUM_ECC_DIGITS] * (!borrow); |
| 275 | borrow = vli_sub(q1, p_product, prime2, NUM_ECC_DIGITS + 1); |
| 276 | vli_cond_set(p_product, p_product, q1, borrow); |
| 277 | p_product[NUM_ECC_DIGITS] = q1[NUM_ECC_DIGITS] * (!borrow); |
| 278 | |
| 279 | for (i = 0; i < NUM_ECC_DIGITS; i++) { |
| 280 | p_result[i] = p_product[i]; |
| 281 | } |
| 282 | } |
| 283 | |
| 284 | /* |
| 285 | * Computes modular exponentiation. |
| 286 | * |
| 287 | * Side-channel countermeasure: algorithm strengthened against timing attack. |
| 288 | */ |
| 289 | static void vli_modExp(uint32_t *p_result, uint32_t *p_base, |
| 290 | uint32_t *p_exp, uint32_t *p_mod, uint32_t *p_barrett) |
| 291 | { |
| 292 | |
| 293 | uint32_t acc[NUM_ECC_DIGITS], tmp[NUM_ECC_DIGITS], product[2 * NUM_ECC_DIGITS]; |
| 294 | uint32_t j; |
| 295 | int32_t i; |
| 296 | |
| 297 | vli_clear(acc); |
| 298 | acc[0] = 1; |
| 299 | |
| 300 | for (i = NUM_ECC_DIGITS - 1; i >= 0; i--) { |
| 301 | for (j = 1 << 31; j > 0; j = j >> 1) { |
| 302 | vli_square(product, acc); |
| 303 | vli_mmod_barrett(acc, product, p_mod, p_barrett); |
| 304 | vli_mult(product, acc, p_base, NUM_ECC_DIGITS); |
| 305 | vli_mmod_barrett(tmp, product, p_mod, p_barrett); |
| 306 | vli_cond_set(acc, tmp, acc, j & p_exp[i]); |
| 307 | } |
| 308 | } |
| 309 | |
| 310 | vli_set(p_result, acc); |
| 311 | } |
| 312 | |
| 313 | /* Conversion from Affine coordinates to Jacobi coordinates. */ |
| 314 | static void EccPoint_fromAffine(EccPointJacobi *p_point_jacobi, |
| 315 | EccPoint *p_point) { |
| 316 | |
| 317 | vli_set(p_point_jacobi->X, p_point->x); |
| 318 | vli_set(p_point_jacobi->Y, p_point->y); |
| 319 | vli_clear(p_point_jacobi->Z); |
| 320 | p_point_jacobi->Z[0] = 1; |
| 321 | } |
| 322 | |
| 323 | /* |
| 324 | * Elliptic curve point doubling in Jacobi coordinates: P = P + P. |
| 325 | * |
| 326 | * Requires 4 squares and 4 multiplications. |
| 327 | */ |
| 328 | static void EccPoint_double(EccPointJacobi *P) |
| 329 | { |
| 330 | |
| 331 | uint32_t m[NUM_ECC_DIGITS], s[NUM_ECC_DIGITS], t[NUM_ECC_DIGITS]; |
| 332 | |
| 333 | vli_modSquare_fast(t, P->Z); |
| 334 | vli_modSub(m, P->X, t, curve_p); |
| 335 | vli_modAdd(s, P->X, t, curve_p); |
| 336 | vli_modMult_fast(m, m, s); |
| 337 | vli_modAdd(s, m, m, curve_p); |
| 338 | vli_modAdd(m, s, m, curve_p); /* m = 3X^2 - 3Z^4 */ |
| 339 | vli_modSquare_fast(t, P->Y); |
| 340 | vli_modMult_fast(s, P->X, t); |
| 341 | vli_modAdd(s, s, s, curve_p); |
| 342 | vli_modAdd(s, s, s, curve_p); /* s = 4XY^2 */ |
| 343 | vli_modMult_fast(P->Z, P->Y, P->Z); |
| 344 | vli_modAdd(P->Z, P->Z, P->Z, curve_p); /* Z' = 2YZ */ |
| 345 | vli_modSquare_fast(P->X, m); |
| 346 | vli_modSub(P->X, P->X, s, curve_p); |
| 347 | vli_modSub(P->X, P->X, s, curve_p); /* X' = m^2 - 2s */ |
| 348 | vli_modSquare_fast(P->Y, t); |
| 349 | vli_modAdd(P->Y, P->Y, P->Y, curve_p); |
| 350 | vli_modAdd(P->Y, P->Y, P->Y, curve_p); |
| 351 | vli_modAdd(P->Y, P->Y, P->Y, curve_p); |
| 352 | vli_modSub(t, s, P->X, curve_p); |
| 353 | vli_modMult_fast(t, t, m); |
| 354 | vli_modSub(P->Y, t, P->Y, curve_p); /* Y' = m(s - X') - 8Y^4 */ |
| 355 | |
| 356 | } |
| 357 | |
| 358 | /* Copy input to target. */ |
| 359 | static void EccPointJacobi_set(EccPointJacobi *target, EccPointJacobi *input) |
| 360 | { |
| 361 | vli_set(target->X, input->X); |
| 362 | vli_set(target->Y, input->Y); |
| 363 | vli_set(target->Z, input->Z); |
| 364 | } |
| 365 | |
| 366 | /* ------ Externally visible functions (see header file for comments): ------ */ |
| 367 | |
| 368 | void vli_set(uint32_t *p_dest, uint32_t *p_src) |
| 369 | { |
| 370 | |
| 371 | uint32_t i; |
| 372 | |
| 373 | for (i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 374 | p_dest[i] = p_src[i]; |
| 375 | } |
| 376 | } |
| 377 | |
| 378 | int32_t vli_cmp(uint32_t *p_left, uint32_t *p_right, int32_t word_size) |
| 379 | { |
| 380 | |
| 381 | int32_t i, cmp = 0; |
| 382 | |
| 383 | for (i = word_size-1; i >= 0; --i) { |
| 384 | cmp |= ((p_left[i] > p_right[i]) - (p_left[i] < p_right[i])) * (!cmp); |
| 385 | } |
| 386 | |
| 387 | return cmp; |
| 388 | } |
| 389 | |
| 390 | uint32_t vli_sub(uint32_t *p_result, uint32_t *p_left, uint32_t *p_right, |
| 391 | uint32_t word_size) |
| 392 | { |
| 393 | |
| 394 | uint32_t l_borrow = 0; |
| 395 | |
| 396 | for (uint32_t i = 0; i < word_size; ++i) { |
| 397 | uint32_t l_diff = p_left[i] - p_right[i] - l_borrow; |
| 398 | |
| 399 | l_borrow = (l_diff > p_left[i]) | ((l_diff == p_left[i]) && l_borrow); |
| 400 | p_result[i] = l_diff; |
| 401 | } |
| 402 | |
| 403 | return l_borrow; |
| 404 | } |
| 405 | |
| 406 | void vli_cond_set(uint32_t *output, uint32_t *p_true, uint32_t *p_false, |
| 407 | uint32_t cond) |
| 408 | { |
| 409 | uint32_t i; |
| 410 | |
| 411 | cond = (!cond); |
| 412 | |
| 413 | for (i = 0; i < NUM_ECC_DIGITS; i++) { |
| 414 | output[i] = (p_true[i]*(!cond)) | (p_false[i]*cond); |
| 415 | } |
| 416 | } |
| 417 | |
| 418 | void vli_modAdd(uint32_t *p_result, uint32_t *p_left, uint32_t *p_right, |
| 419 | uint32_t *p_mod) |
| 420 | { |
| 421 | uint32_t l_carry = vli_add(p_result, p_left, p_right); |
| 422 | uint32_t p_temp[NUM_ECC_DIGITS]; |
| 423 | |
| 424 | l_carry = l_carry == vli_sub(p_temp, p_result, p_mod, NUM_ECC_DIGITS); |
| 425 | vli_cond_set(p_result, p_temp, p_result, l_carry); |
| 426 | } |
| 427 | |
| 428 | void vli_modSub(uint32_t *p_result, uint32_t *p_left, uint32_t *p_right, |
| 429 | uint32_t *p_mod) |
| 430 | { |
| 431 | uint32_t l_borrow = vli_sub(p_result, p_left, p_right, NUM_ECC_DIGITS); |
| 432 | uint32_t p_temp[NUM_ECC_DIGITS]; |
| 433 | |
| 434 | vli_add(p_temp, p_result, p_mod); |
| 435 | vli_cond_set(p_result, p_temp, p_result, l_borrow); |
| 436 | } |
| 437 | |
| 438 | void vli_modMult_fast(uint32_t *p_result, uint32_t *p_left, |
| 439 | uint32_t *p_right) |
| 440 | { |
| 441 | uint32_t l_product[2 * NUM_ECC_DIGITS]; |
| 442 | |
| 443 | vli_mult(l_product, p_left, p_right, NUM_ECC_DIGITS); |
| 444 | vli_mmod_barrett(p_result, l_product, curve_p, curve_pb); |
| 445 | } |
| 446 | |
| 447 | void vli_modSquare_fast(uint32_t *p_result, uint32_t *p_left) |
| 448 | { |
| 449 | uint32_t l_product[2 * NUM_ECC_DIGITS]; |
| 450 | |
| 451 | vli_square(l_product, p_left); |
| 452 | vli_mmod_barrett(p_result, l_product, curve_p, curve_pb); |
| 453 | } |
| 454 | |
| 455 | void vli_modMult(uint32_t *p_result, uint32_t *p_left, uint32_t *p_right, |
| 456 | uint32_t *p_mod, uint32_t *p_barrett) |
| 457 | { |
| 458 | |
| 459 | uint32_t l_product[2 * NUM_ECC_DIGITS]; |
| 460 | |
| 461 | vli_mult(l_product, p_left, p_right, NUM_ECC_DIGITS); |
| 462 | vli_mmod_barrett(p_result, l_product, p_mod, p_barrett); |
| 463 | } |
| 464 | |
| 465 | void vli_modInv(uint32_t *p_result, uint32_t *p_input, uint32_t *p_mod, |
| 466 | uint32_t *p_barrett) |
| 467 | { |
| 468 | uint32_t p_power[NUM_ECC_DIGITS]; |
| 469 | |
| 470 | vli_set(p_power, p_mod); |
| 471 | p_power[0] -= 2; |
| 472 | vli_modExp(p_result, p_input, p_power, p_mod, p_barrett); |
| 473 | } |
| 474 | |
| 475 | uint32_t EccPoint_isZero(EccPoint *p_point) |
| 476 | { |
| 477 | return (vli_isZero(p_point->x) && vli_isZero(p_point->y)); |
| 478 | } |
| 479 | |
| 480 | uint32_t EccPointJacobi_isZero(EccPointJacobi *p_point_jacobi) |
| 481 | { |
| 482 | return vli_isZero(p_point_jacobi->Z); |
| 483 | } |
| 484 | |
| 485 | void EccPoint_toAffine(EccPoint *p_point, EccPointJacobi *p_point_jacobi) |
| 486 | { |
| 487 | |
| 488 | if (vli_isZero(p_point_jacobi->Z)) { |
| 489 | vli_clear(p_point->x); |
| 490 | vli_clear(p_point->y); |
| 491 | return; |
| 492 | } |
| 493 | |
| 494 | uint32_t z[NUM_ECC_DIGITS]; |
| 495 | |
| 496 | vli_set(z, p_point_jacobi->Z); |
| 497 | vli_modInv(z, z, curve_p, curve_pb); |
| 498 | vli_modSquare_fast(p_point->x, z); |
| 499 | vli_modMult_fast(p_point->y, p_point->x, z); |
| 500 | vli_modMult_fast(p_point->x, p_point->x, p_point_jacobi->X); |
| 501 | vli_modMult_fast(p_point->y, p_point->y, p_point_jacobi->Y); |
| 502 | } |
| 503 | |
| 504 | void EccPoint_add(EccPointJacobi *P1, EccPointJacobi *P2) |
| 505 | { |
| 506 | |
| 507 | uint32_t s1[NUM_ECC_DIGITS], u1[NUM_ECC_DIGITS], t[NUM_ECC_DIGITS]; |
| 508 | uint32_t h[NUM_ECC_DIGITS], r[NUM_ECC_DIGITS]; |
| 509 | |
| 510 | vli_modSquare_fast(r, P1->Z); |
| 511 | vli_modSquare_fast(s1, P2->Z); |
| 512 | vli_modMult_fast(u1, P1->X, s1); /* u1 = X1 Z2^2 */ |
| 513 | vli_modMult_fast(h, P2->X, r); |
| 514 | vli_modMult_fast(s1, P1->Y, s1); |
| 515 | vli_modMult_fast(s1, s1, P2->Z); /* s1 = Y1 Z2^3 */ |
| 516 | vli_modMult_fast(r, P2->Y, r); |
| 517 | vli_modMult_fast(r, r, P1->Z); |
| 518 | vli_modSub(h, h, u1, curve_p); /* h = X2 Z1^2 - u1 */ |
| 519 | vli_modSub(r, r, s1, curve_p); /* r = Y2 Z1^3 - s1 */ |
| 520 | |
| 521 | if (vli_isZero(h)) { |
| 522 | if (vli_isZero(r)) { |
| 523 | /* P1 = P2 */ |
| 524 | EccPoint_double(P1); |
| 525 | return; |
| 526 | } |
| 527 | /* point at infinity */ |
| 528 | vli_clear(P1->Z); |
| 529 | return; |
| 530 | } |
| 531 | |
| 532 | vli_modMult_fast(P1->Z, P1->Z, P2->Z); |
| 533 | vli_modMult_fast(P1->Z, P1->Z, h); /* Z3 = h Z1 Z2 */ |
| 534 | vli_modSquare_fast(t, h); |
| 535 | vli_modMult_fast(h, t, h); |
| 536 | vli_modMult_fast(u1, u1, t); |
| 537 | vli_modSquare_fast(P1->X, r); |
| 538 | vli_modSub(P1->X, P1->X, h, curve_p); |
| 539 | vli_modSub(P1->X, P1->X, u1, curve_p); |
| 540 | vli_modSub(P1->X, P1->X, u1, curve_p); /* X3 = r^2 - h^3 - 2 u1 h^2 */ |
| 541 | vli_modMult_fast(t, s1, h); |
| 542 | vli_modSub(P1->Y, u1, P1->X, curve_p); |
| 543 | vli_modMult_fast(P1->Y, P1->Y, r); |
| 544 | vli_modSub(P1->Y, P1->Y, t, curve_p); /* Y3 = r(u1 h^2 - X3) - s1 h^3 */ |
| 545 | } |
| 546 | |
| 547 | /* |
| 548 | * Elliptic curve scalar multiplication with result in Jacobi coordinates: |
| 549 | * |
| 550 | * p_result = p_scalar * p_point. |
| 551 | */ |
| 552 | void EccPoint_mult_safe(EccPointJacobi *p_result, EccPoint *p_point, uint32_t *p_scalar) |
| 553 | { |
| 554 | |
| 555 | int32_t i; |
| 556 | uint32_t bit; |
| 557 | EccPointJacobi p_point_jacobi, p_tmp; |
| 558 | |
| 559 | EccPoint_fromAffine(p_result, p_point); |
| 560 | EccPoint_fromAffine(&p_point_jacobi, p_point); |
| 561 | |
| 562 | for (i = vli_numBits(p_scalar) - 2; i >= 0; i--) { |
| 563 | EccPoint_double(p_result); |
| 564 | EccPointJacobi_set(&p_tmp, p_result); |
| 565 | EccPoint_add(&p_tmp, &p_point_jacobi); |
| 566 | bit = vli_testBit(p_scalar, i); |
| 567 | vli_cond_set(p_result->X, p_tmp.X, p_result->X, bit); |
| 568 | vli_cond_set(p_result->Y, p_tmp.Y, p_result->Y, bit); |
| 569 | vli_cond_set(p_result->Z, p_tmp.Z, p_result->Z, bit); |
| 570 | } |
| 571 | } |
| 572 | |
| 573 | /* Ellptic curve scalar multiplication with result in Jacobi coordinates */ |
| 574 | /* p_result = p_scalar * p_point */ |
| 575 | void EccPoint_mult_unsafe(EccPointJacobi *p_result, EccPoint *p_point, uint32_t *p_scalar) |
| 576 | { |
| 577 | int i; |
| 578 | EccPointJacobi p_point_jacobi; |
| 579 | EccPoint_fromAffine(p_result, p_point); |
| 580 | EccPoint_fromAffine(&p_point_jacobi, p_point); |
| 581 | |
| 582 | for(i = vli_numBits(p_scalar) - 2; i >= 0; i--) |
| 583 | { |
| 584 | EccPoint_double(p_result); |
| 585 | if (vli_testBit(p_scalar, i)) |
| 586 | { |
| 587 | EccPoint_add(p_result, &p_point_jacobi); |
| 588 | } |
| 589 | } |
| 590 | } |
| 591 | |
| 592 | /* -------- Conversions between big endian and little endian: -------- */ |
| 593 | |
| 594 | void ecc_bytes2native(uint32_t p_native[NUM_ECC_DIGITS], |
| 595 | uint8_t p_bytes[NUM_ECC_DIGITS * 4]) |
| 596 | { |
| 597 | |
| 598 | uint32_t i; |
| 599 | |
| 600 | for (i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 601 | uint8_t *p_digit = p_bytes + 4 * (NUM_ECC_DIGITS - 1 - i); |
| 602 | |
| 603 | p_native[i] = ((uint32_t)p_digit[0] << 24) | |
| 604 | ((uint32_t)p_digit[1] << 16) | |
| 605 | ((uint32_t)p_digit[2] << 8) | |
| 606 | (uint32_t)p_digit[3]; |
| 607 | } |
| 608 | } |
| 609 | |
| 610 | void ecc_native2bytes(uint8_t p_bytes[NUM_ECC_DIGITS * 4], |
| 611 | uint32_t p_native[NUM_ECC_DIGITS]) |
| 612 | { |
| 613 | |
| 614 | uint32_t i; |
| 615 | |
| 616 | for (i = 0; i < NUM_ECC_DIGITS; ++i) { |
| 617 | uint8_t *p_digit = p_bytes + 4 * (NUM_ECC_DIGITS - 1 - i); |
| 618 | |
| 619 | p_digit[0] = p_native[i] >> 24; |
| 620 | p_digit[1] = p_native[i] >> 16; |
| 621 | p_digit[2] = p_native[i] >> 8; |
| 622 | p_digit[3] = p_native[i]; |
| 623 | } |
| 624 | } |
| 625 | |